Natalie Pariente-Cohen,
Michal Weitman,
Nassdyuk Tania,
Dan T. Major,
Hugo E. Gottlieb,
Shmaryahu Hoz and
Abraham Nudelman*
Chemistry Department, Bar-Ilan University, Ramat Gan 52900, Israel. E-mail: nudelman@biu.ac.il
First published on 24th February 2015
Attempted acylation of the anticancer agent hydroxyurea (HU) with acyl chlorides or anhydrides led to acylation on the NH group rather than on the OH. The structures of the products were confirmed by 15N-HMBC NMR. An analogous reaction conducted with hydroxamic acids (RCONHOH) or N-hydroxycarbamates (ROCONHOH) led to acylation on the OH. Surprisingly, despite the established affinity of phosphorous to O, phosphorylation of HU also took place at the NH group instead of the OH. These results are rationalized based on the different dominant resonance structures of HU, the hydroxamic acids or the N-hydroxycarbamates.
Although a large number of hydroxamic acid derivatives have been reported in the literature,7 and many patents claim various uses for these compounds, only a few compounds that possess hydroxamic acid groups (R–CONHOH) are currently in clinical use, including: vorinostat, as a HDAC inhibitor, and HU as an anticancer agent. Thus, we intended to prepare and study the esters of HU connected to acidic HDAC inhibitors. A literature search of acylated HU derivatives revealed that very few publications claim the synthesis of HU esters, described as being synthesized by O-acylation of HU8 or via the reaction of isocyanates with hydroxylamine.9 Whereas Exner reported10 that the benzoylation of HU with benzoic anhydride led to the benzoate ester 1 (Scheme 1), in our studies we found that the acylation took place easily on the secondary nitrogen rather than on the hydroxyl group. Herein we report on various products obtained from the reaction of activated acids with HU. To prove our claim that the acylation takes place preferentially on the NH group, we repeated Exner's procedure10 and found that the obtained product was the amide 2 (ref. 11) rather than the claimed ester 1, or the acylated enolic form of HU 3 (Scheme 1).
Since the 1H NMR spectrum of the product is inconclusive as to the site of acylation, whether it takes place on the N or on either of the O's, and in the 13C NMR the chemical shift of the carbonyl of the acyl group does not readily distinguish between a CO that belongs to an ester or to an amide, we resorted to determine the 15N-HMBC NMR spectrum of the product. This experiment, based on the 15N-1H coupling identifies unambiguously, NH or NH2 groups, as the 15N signal of these is split in the 1H dimension into a doublet (J = ca. 90 Hz) due to the one-bond coupling in the presence of –NH– or –NH2. When the H is not directly attached to the N, but is found 2 or three atoms away, the longer-range coupling constants (2J and 3J) are much smaller, and even if their correlation peaks appear, they are not visibly split. Furthermore, the chemical shift of the N in NH2 groups in ureas is usually found at ∼–300 ppm, that of amides NCOR at ∼–200 ppm, and that of oximes
N–OH at ∼–50 ppm.12 By examination of the NMR spectrum, we can identify which of the protons on a heteroatom are connected to a N atom and by exclusion, those that are connected to O. In the spectrum of the benzoylated product of HU the following observations are made: (a) the 15N-HMBC showed an N signal at −305.8 ppm, which was attached through one bond to the two H's at 6.36 ppm, indicating the presence of an amide –CONH2; (b) a second N that appears at ∼−200 ppm indicative of an NCOR; (c) no N is found to have chemical shift of ∼−50 ppm; and (e) the other 1H on a heteroatom (δ 9.24) is not connected to a N and is therefore assigned to the OH group. Based on these observations, the assignment of the structure of the product is that of compound 2 (Fig. 1).
![]() | ||
Fig. 1 1Hx15N NMR correlation spectrum of compound 2, the N-benzoylated product of HU; inset: 1H-1D spectrum. |
When analogous acylations of HU were carried out with valproyl chloride the main product 4 was again found to be that of N-acylation. In addition, the acylation led to the formation of a small amount of the N,O-bis-valproylated hydroxylamine 5, the formation of which may be accounted for by decomposition of an N,O-bis-valproylated HU intermediate with concomitant release of HNCO, or by initial breakdown of 4 to give the N-acylated hydroxylamine, which underwent further O-acylation with another equivalent of the valproyl chloride to give compound 5 (Scheme 2). A similar reaction was reported by Exner10 involving the exclusive formation of the analog of 5 (R = Ph)13 upon treatment of HU with benzoyl chloride (Fig. 2).
![]() | ||
Fig. 2 1Hx15N NMR correlation spectrum of compound 4, the N-valproylated product of HU; inset: 1H-1D spectrum. |
Since HU possess a –CONHOH group, analogous to that found in hydroxamic acids, we examined the reactions of N-hydroxypivalamide 7 (ref. 15) and the N-hydroxycarbamate 10 (ref. 16) with valproyl chloride. In these cases the products of O-acylation 8 and 11 formed readily, and no N-acylated products 9 or 12 were detected. This outcome shows that HU is an unusual hydroxamic acid and the presence of the NH2 group instead of alkyl or alkoxy groups caused the HU to react differently (Scheme 3).
In an attempt to understand the different behavior of HU toward acylation in comparison to other hydroxamic acid analogs we compared the resonance structures expected for these compounds. In the case of HU, we suggest that the primary resonance takes place between the unshared electrons on the NH2 nitrogen and the oxygen of the carbonyl, whereas in the hydroxamic acid and the N-hydroxycarbamate the main resonance takes place between the unshared electrons on the NH nitrogen and the oxygen of the carbonyl (Scheme 4). Thus, in the case of the HU the unshared electrons on the NH group are readily available for nucleophilic attack leading to the found N-hydroxyimide. Further support for this suggestion is based on the fact that X-ray crystallography of HU reveals that the length of the C–NH2 bond is 1.328 Å whereas that of the C–NHOH is 1.347 Å.17 This length difference may indicate that the C–NH2 contributes the dominant resonance structure and might have more sp2 hybridization, making the N of the –NHOH group more basic-nucleophilic leading to the observed N-acylation. This observation, however does not clarify why the acylation takes place on the N and not on the O of the NHOH, and moreover it does not rationalize why the phosphorylation also takes place on the N rather than on the O, despite the well-established affinity of P to O. Whereas in the hydroxamic acid and the N-hydroxycarbamate the electron concentration of the NH is diminished, leading to the acylation of the OH groups.
![]() | ||
Scheme 4 Some suggested resonance structures and the obtained acylation products of HU, a hydroxamic acid and an N-hydroxycarbamate. |
An additional cause for the preference of NH over the NH2 as the nucleophilic site comes from the α-effect.18 This effect is observed in cases where a lone pair, carrying atom resides α to the nucleophilic atom. The anion of hydrogen peroxide is a classic example of this group of nucleophiles, which exhibit an enhanced nucleophilicity mainly towards compounds with a low lying LUMO.18e Hence, because of the neighboring oxygen atom, the NH will be more nucleophilic than the NH2 group.
In view of the unexpected course of HU acylation described above, we proceeded with an attempted phosphorylation of HU. In this case, based on the well established, high-affinity of phosphorous to oxygen, it was expected that the phosphorylation would take place on the oxygen of HU. Surprisingly, when HU was reacted with diethyl chlorophosphate the phosphorylation again took place on the NH and not on the OH to give the corresponding diethyl phosphoramidate 13 instead of the phosphates 14 or 15 (Scheme 5). The structure of the isolated product 13 was also established by its 15N-HMBC NMR and by the chemical shifts in the 15N NMR spectrum, where the N's have chemical shifts of ∼−300 ppm and ∼−200 ppm, and not ∼−50 ppm as would have been expected had the compound contain an N–OH group as that shown in structure 15 (see experimental). It appears that based on the above suggested resonance argument, the N in the NH group is of sufficiently nucleophilic character that it overcomes the well-established affinity of O to P, leading to the N-phosphorylated product.
1H NMR (300 MHz, acetone-D6): δ: 6.36 (bs, 2H), 7.56 (t, J = 15.93 Hz, 2H), 7.71 (t, J = 15.93 Hz, 2H), 8.10 (d, J = 6.37 Hz, 2H), 9.24 (bs, 1H).
13C NMR (75 MHz, acetone-D6): δ: 128.68, 129.60, 130.48, 134.74, 160.48, 166.19.
15N NMR (indirectly from HMBC spectrum): δ: −305.8 (NH2).
MS (ES+) m/z 181 (MH+, 100); 203 ([M + Na]+, 96).
Anal. calcd for C8H8N2O3: C, 53.33; H, 4.48; N, 15.55; O, 26.64. found: C, 53.48; H, 4.38; N, 15.35; O, 26.14%.
1H NMR of compound 4 (700 MHz, CDCl3): δ: 0.91 (t, J = 7 Hz, 6H), 1.33 (m, 4H), 1.51 (m, 2H), 1.66 (m, 2H), 2.52 (m, 1H), 5.85 (bs, 2H), 9.14 (bs, 1H).
13C NMR (176 MHz, CDCl3): δ: 13.90, 20.54, 34.40, 43.40, 160.09, 174.66.
15N NMR (obtained indirectly from the HMBC spectrum): δ: −221 (N–OH), −303 (NH2).
MS (ES+) m/z 203 (MH+, 100); 225 ([M + Na]+, 96). HRMS calcd for C9H18N2O3Na (M+, Na+) 225.1222, found 225.1215.
1H NMR of compound 5 (700 MHz, CDCl3): δ: 0.91 (m, 12H), 1.28–1.45 (m, 10H), 1.51 (m, 2H), 1.67 (m, 4H), 2.15 (m, 1H), 2.57 (m, 1H), 8.95 (s, 1H).
13C NMR (176 MHz, CDCl3): δ: 13.94, 14.05, 20.51, 20.65, 34.43, 34.86, 43.19, 44.24, 174.40, 174.98.
15N NMR (obtained indirectly from the HMBC spectrum): δ: −208.8.
MS (ES+) m/z 286 (MH+, 100); 308 ([M + Na]+, 9).
HRMS (APPI+) calcd for C16H31NO3Na (M+, Na+) 308.2202, found 308.2209.
1H NMR (700 MHz, CDCl3): δ: 0.91 (t, J = 7 Hz, 6H), 1.28 (bs, 9H), 1.38 (m, 4H), 1.51 (m, 2H), 1.68 (m, 2H), 2.57 (m, 1H), 9.04 (s, 1H).
13C NMR (176 MHz, CDCl3): δ: 13.94, 20.49, 27.21, 34.42, 38.68, 43.22, 175.17, 176.85.
15N NMR (obtained indirectly from the HMBC spectrum, 700 MHz, CDCl3): δ: −212.2.
MS (ES+) m/z 244 (MH+, 100), 266 ([M + Na]+, 28.86).
Anal. calcd for C13H25NO3: C, 64.16; H, 10.36; N, 5.76; found: C, 63.46; H, 10.75; N, 5.81%.
1H NMR (700 MHz, CDCl3): δ: 0.84 (m, 6H), 1.28 (m, 4H), 1.40 (m, 9H), 1.60 (s, 2H), 2.46 (m, 1H), 7.98 (s, 1H).
13C NMR (176 MHz, CDCl3): δ: 13.74, 20.30, 27.87, 34.19, 42.97, 82.74, 155.54, 175.64.
15N NMR (obtained indirectly from the HMBC spectrum, 700 MHz, acetone-D6): δ: −240.0.
MS (ES+) m/z 282 ([M + Na]+, 40).
HRMS (APPI−) calcd for C13H25NO4 (M−) 258.1705, found 258.1718.
1H NMR (700 MHz, CDCl3): δ: 1.37 (t, J = 8 Hz, 6H), 4.26 (m, 4H), 6.22 (s, 2H), 8.98 (s, 1H).
13C NMR (176 MHz, CDCl3): δ: 16.03 (d, J = 6 Hz), 65.72 (d, J = 6 Hz), 161.47 (d, J = 4 Hz).
15N NMR (indirectly from HMBC spectrum): δ: −220.1 (N–OH), −300.8 (NH2CO).
TOF MS (ES+) m/z 235 ([M + Na]+, 100).
31P NMR (400 MHz, CDCl3): δ: 1.16.
Anal. calcd for C5H13N2O5P: C, 28.31; H, 6.18; N, 13.21; found: C, 28.84; H, 6.15; N, 13.06%.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5ra01016k |
This journal is © The Royal Society of Chemistry 2015 |